---
id: video/gd2TycVywv4
type: video
title: 20260326 - From AI Novice to Agentic Thinking For the CAL Brained
summary: This is a webinar with Jeremy Visca on agentic thinking for long-time NAV/CAL and Business Central AL developers. It covers the shift from writing every line to specifying intent and validating outcomes, with tests as the safety net. It also covers managing agent context like an intern's desk, and starting gently with documentation, tests, scaffolding and design critique.
tier: community
language: en
tags:
  - agentic thinking
  - ai adoption
  - mindset shift
  - code generation
  - prompt engineering
  - developer skills
  - maintainability
  - democratization
  - github copilot
  - business central development
  - al development
  - code review
system: development
review:
  state: reviewed
  by: opus
  at: "2026-10-07T23:21:22.673Z"
  flags: []
generated:
  at: "2026-10-07T23:21:22.724Z"
  pipeline: 0.2.0
  prompts:
    extract-video: 1
    summarize-video: 2
  input_hash: d064ea8f6e181f14bb031b8192fa726d4d2c38ee2d7b420de9a14e181c53a7b0
evidence:
  - kind: video
    url: https://www.youtube.com/watch?v=gd2TycVywv4&t=462s
    title: 20260326 - From AI Novice to Agentic Thinking For the CAL Brained
    date: "2026-03-24T21:02:55.000Z"
    commit: null
    t: 462
    quote: We bring change to our customers, and now change has come for us.
  - kind: video
    url: https://www.youtube.com/watch?v=gd2TycVywv4&t=488s
    title: 20260326 - From AI Novice to Agentic Thinking For the CAL Brained
    date: "2026-03-24T21:02:55.000Z"
    commit: null
    t: 488
    quote: you come at things with the I write every line of code. I understand everything that's in here.
  - kind: video
    url: https://www.youtube.com/watch?v=gd2TycVywv4&t=593s
    title: 20260326 - From AI Novice to Agentic Thinking For the CAL Brained
    date: "2026-03-24T21:02:55.000Z"
    commit: null
    t: 593
    quote: In agentic world, the agent might do different stuff every time. That's a very, very big shift.
  - kind: video
    url: https://www.youtube.com/watch?v=gd2TycVywv4&t=605s
    title: 20260326 - From AI Novice to Agentic Thinking For the CAL Brained
    date: "2026-03-24T21:02:55.000Z"
    commit: null
    t: 605
    quote: your intent and your ability to sculpt the intent for the agent to understand is now the new mastery.
  - kind: video
    url: https://www.youtube.com/watch?v=gd2TycVywv4&t=718s
    title: 20260326 - From AI Novice to Agentic Thinking For the CAL Brained
    date: "2026-03-24T21:02:55.000Z"
    commit: null
    t: 718
    quote: matters more than whether or not you're working with ai
links:
  learn: []
  objects: []
  features: []
  topics: []
  localizations: []
  videos: []
  posts: []
  guidelines: []
video_id: gd2TycVywv4
channel: yt-areopa
source_name: Areopa webinars
url: https://www.youtube.com/watch?v=gd2TycVywv4
published_at: "2026-03-24T21:02:55.000Z"
duration_s: 3722
captions: derived
audience:
  - developer
  - decision maker
  - functional consultant
chapters:
  - t: 0
    title: Webinar introduction and housekeeping
  - t: 145
    title: Website updates and upcoming sessions
  - t: 308
    title: Jeremy Visca introduction and background
  - t: 425
    title: Why AI and agentic thinking is hard for developers
  - t: 552
    title: Shift from deterministic code to fuzzy agentic outcomes
  - t: 643
    title: AI as amplifier and research on code maintainability
  - t: 730
    title: Developer skill and transition history
  - t: 794
    title: Agentic era and democratization of development
  - t: 911
    title: Code volume and architectural tradeoffs
  - t: 1062
    title: Testing as a safety net for AI-generated code
  - t: 1243
    title: Mindset shifts for agentic development
  - t: 1407
    title: Applying domain expertise to guide agents and testing practices
  - t: 1780
    title: Learning from community and the master loop pattern
  - t: 2035
    title: Learning and growth through agentic experiences and mental models
  - t: 2238
    title: Context windows, instructions, and model selection
  - t: 2460
    title: Context limits, summarization risks, and agent failures
  - t: 2686
    title: Agentic development as learnable skill and job security concerns
  - t: 2878
    title: Starting gently with documentation and progressive scaffolding
  - t: 3088
    title: Autonomous agent workflows and building management skills
  - t: 3225
    title: Overcoming objections and establishing documentation standards
  - t: 3426
    title: GitHub examples and verification practices
  - t: 3509
    title: Career progression for developers in the AI era
  - t: 3610
    title: Closing remarks and session conclusion
features:
  - name: Agentic development training program
    status: unclear
    t: 414
    verified: false
    status_source: video
  - name: AI-assisted development with GitHub Copilot
    status: unclear
    t: 879
    verified: false
    status_source: video
  - name: LLM-based compilation model
    status: unclear
    t: 794
    verified: false
    status_source: video
  - name: Agentic code testing requirements
    status: unclear
    t: 1080
    verified: false
    status_source: video
  - name: BC Code Intel MCP
    status: unclear
    t: 1541
    verified: false
    status_source: video
  - name: Agentic development mindset shift
    status: unclear
    t: 1307
    verified: false
    status_source: video
  - name: Domain expertise validation for agent-generated code
    status: unclear
    t: 1590
    verified: false
    status_source: video
  - name: Result-based evaluation replacing line-by-line code review
    status: unclear
    t: 1724
    verified: false
    status_source: video
  - name: Master loop pattern for agent development
    status: unclear
    t: 1900
    verified: false
    status_source: video
  - name: BC Code Intel repository
    status: unclear
    t: 1924
    verified: false
    status_source: video
  - name: Calibrated confidence in agent code
    status: unclear
    t: 1833
    verified: false
    status_source: video
  - name: Intent steering in agent development
    status: unclear
    t: 1890
    verified: false
    status_source: video
  - name: Using agents to explain developer commits
    status: unclear
    t: 2068
    verified: false
    status_source: video
  - name: Managing multiple concurrent agents
    status: unclear
    t: 2104
    verified: false
    status_source: video
  - name: Agent mental model as overenthusiastic intern
    status: unclear
    t: 2128
    verified: false
    status_source: video
  - name: Context window limitations and token limits
    status: unclear
    t: 2198
    verified: false
    status_source: video
  - name: Instructions and custom agent configuration
    status: unclear
    t: 2238
    verified: false
    status_source: video
  - name: LLM input-output architecture for agents
    status: unclear
    t: 2287
    verified: false
    status_source: video
  - name: Model selection and cost-performance tradeoff
    status: unclear
    t: 2341
    verified: false
    status_source: video
  - name: Signal-to-noise ratio in agent context
    status: unclear
    t: 2398
    verified: false
    status_source: video
  - name: Agent context optimization
    status: unclear
    t: 2375
    verified: false
    status_source: video
  - name: Agent session reset for fresh state
    status: unclear
    t: 2496
    verified: false
    status_source: video
  - name: Context window compaction and summarization
    status: unclear
    t: 2518
    verified: false
    status_source: video
  - name: Specific task decomposition for agents
    status: unclear
    t: 2661
    verified: false
    status_source: video
  - name: Agentic development as learnable skill
    status: unclear
    t: 2733
    verified: false
    status_source: video
  - name: Agents on legacy and messy code
    status: unclear
    t: 2769
    verified: false
    status_source: video
  - name: Documentation assistance via agents
    status: unclear
    t: 2878
    verified: false
    status_source: video
  - name: Test generation and scenario expansion
    status: unclear
    t: 2923
    verified: false
    status_source: video
  - name: Scaffolding and table design via agents
    status: unclear
    t: 2987
    verified: false
    status_source: video
  - name: Design validation and assumption challenging
    status: unclear
    t: 3052
    verified: false
    status_source: video
  - name: Autonomous agent compilation and testing
    status: unclear
    t: 3088
    verified: false
    status_source: video
  - name: Agent-driven code improvement through documentation and quality verification
    status: unclear
    t: 3140
    verified: false
    status_source: video
  - name: XML documentation for code objects improving AI understanding
    status: unclear
    t: 3345
    verified: false
    status_source: video
  - name: AL guidelines best practices resource
    status: unclear
    t: 3260
    verified: false
    status_source: video
objects_mentioned:
  - other BC Code Intel MCP
  - other BC Code Intel
  - other MCP servers
  - other telemetry buddy
  - other object ID Ninja
quotes:
  - t: 462
    text: We bring change to our customers, and now change has come for us.
    check: exact
  - t: 488
    text: you come at things with the I write every line of code. I understand everything that's in here.
    check: exact
  - t: 593
    text: In agentic world, the agent might do different stuff every time. That's a very, very big shift.
    check: exact
  - t: 605
    text: your intent and your ability to sculpt the intent for the agent to understand is now the new mastery.
    check: exact
  - t: 718
    text: matters more than whether or not you're working with ai
    check: fuzzy
---

# 20260326 - From AI Novice to Agentic Thinking For the CAL Brained

> This is a webinar with Jeremy Visca on agentic thinking for long-time NAV/CAL and Business Central AL developers. It covers the shift from writing every line to specifying intent and validating outcomes, with tests as the safety net. It also covers managing agent context like an intern's desk, and starting gently with documentation, tests, scaffolding and design critique.

[Watch on YouTube](https://www.youtube.com/watch?v=gd2TycVywv4) · Areopa webinars · 2026-03-24 · 1:02:02 · tier community · reviewed (checked by Opus)

## Overview

Jeremy Visca explains why AI and agentic development is hard for developers who are used to writing and understanding every line of code. He describes the shift from deterministic code to fuzzy outcomes, where the LLM acts as the compiler and the context and prompts are the product. He shows a customer example where GitHub Copilot was used to create a Power BI API endpoint in minutes.

The session then covers practical habits: strong tests and performance tests, domain-expertise checks on agent output, the master loop of asking, correcting and asking what was learned, and keeping context concise. It closes with ways to start gently (documentation, test expansion, scaffolding, design critique), use on legacy code, and career progression for developers.

## Key points

- Value shifts from writing lines to specifying intent clearly and validating outcomes. The speaker calls this trading certainty for velocity, which is hard in an industry that needs correct accounting outcomes.
- Testing is the safety net for AI-generated code. Agents can be taught to publish to Docker or a sandbox and run the tests. However, they may not think to write performance tests, and they sometimes change tests to match the code.
- Senior developers should apply BC domain knowledge to agent output. For example, when a field is added to a sales order, check the credit memo, invoice and shipment, reverse processes and register writes.
- Context windows are often about 128,000 to 160,000 tokens. Each rule in the BC Code Intel knowledge repo is kept under 100 lines because the agent might not read past that. Separately, when the window fills up, summarization (compaction) can drop critical constraints. Start new sessions often.
- The master loop: ask the agent to do the work, correct what it got wrong, then ask what it could have done differently next time.
- Good starting points are documentation (READMEs, commit explainers, XML documentation with summary tags), extra test scenarios, table and codeunit scaffolding, and design critique from several stakeholder views.
- Agents can be used on legacy code maintained since NAV 4.0. The gains come from parallel work and quality uplift (documentation, tests) rather than raw speed.

## Chapters

- [0:00](https://www.youtube.com/watch?v=gd2TycVywv4&t=0s) Webinar introduction and housekeeping
- [2:25](https://www.youtube.com/watch?v=gd2TycVywv4&t=145s) Website updates and upcoming sessions
- [5:08](https://www.youtube.com/watch?v=gd2TycVywv4&t=308s) Jeremy Visca introduction and background
- [7:05](https://www.youtube.com/watch?v=gd2TycVywv4&t=425s) Why AI and agentic thinking is hard for developers
- [9:12](https://www.youtube.com/watch?v=gd2TycVywv4&t=552s) Shift from deterministic code to fuzzy agentic outcomes
- [10:43](https://www.youtube.com/watch?v=gd2TycVywv4&t=643s) AI as amplifier and research on code maintainability
- [12:10](https://www.youtube.com/watch?v=gd2TycVywv4&t=730s) Developer skill and transition history
- [13:14](https://www.youtube.com/watch?v=gd2TycVywv4&t=794s) Agentic era and democratization of development
- [15:11](https://www.youtube.com/watch?v=gd2TycVywv4&t=911s) Code volume and architectural tradeoffs
- [17:42](https://www.youtube.com/watch?v=gd2TycVywv4&t=1062s) Testing as a safety net for AI-generated code
- [20:43](https://www.youtube.com/watch?v=gd2TycVywv4&t=1243s) Mindset shifts for agentic development
- [23:27](https://www.youtube.com/watch?v=gd2TycVywv4&t=1407s) Applying domain expertise to guide agents and testing practices
- [29:40](https://www.youtube.com/watch?v=gd2TycVywv4&t=1780s) Learning from community and the master loop pattern
- [33:55](https://www.youtube.com/watch?v=gd2TycVywv4&t=2035s) Learning and growth through agentic experiences and mental models
- [37:18](https://www.youtube.com/watch?v=gd2TycVywv4&t=2238s) Context windows, instructions, and model selection
- [41:00](https://www.youtube.com/watch?v=gd2TycVywv4&t=2460s) Context limits, summarization risks, and agent failures
- [44:46](https://www.youtube.com/watch?v=gd2TycVywv4&t=2686s) Agentic development as learnable skill and job security concerns
- [47:58](https://www.youtube.com/watch?v=gd2TycVywv4&t=2878s) Starting gently with documentation and progressive scaffolding
- [51:28](https://www.youtube.com/watch?v=gd2TycVywv4&t=3088s) Autonomous agent workflows and building management skills
- [53:45](https://www.youtube.com/watch?v=gd2TycVywv4&t=3225s) Overcoming objections and establishing documentation standards
- [57:06](https://www.youtube.com/watch?v=gd2TycVywv4&t=3426s) GitHub examples and verification practices
- [58:29](https://www.youtube.com/watch?v=gd2TycVywv4&t=3509s) Career progression for developers in the AI era
- [1:00:10](https://www.youtube.com/watch?v=gd2TycVywv4&t=3610s) Closing remarks and session conclusion

## Features

| Feature | Status | At |
|---|---|---|
| Agentic development training program | status not stated | [6:54](https://www.youtube.com/watch?v=gd2TycVywv4&t=414s) |
| AI-assisted development with GitHub Copilot | status not stated, demoed | [14:39](https://www.youtube.com/watch?v=gd2TycVywv4&t=879s) |
| LLM-based compilation model | status not stated | [13:14](https://www.youtube.com/watch?v=gd2TycVywv4&t=794s) |
| Agentic code testing requirements | status not stated | [18:00](https://www.youtube.com/watch?v=gd2TycVywv4&t=1080s) |
| BC Code Intel MCP | status not stated | [25:41](https://www.youtube.com/watch?v=gd2TycVywv4&t=1541s) |
| Agentic development mindset shift | status not stated | [21:47](https://www.youtube.com/watch?v=gd2TycVywv4&t=1307s) |
| Domain expertise validation for agent-generated code | status not stated | [26:30](https://www.youtube.com/watch?v=gd2TycVywv4&t=1590s) |
| Result-based evaluation replacing line-by-line code review | status not stated | [28:44](https://www.youtube.com/watch?v=gd2TycVywv4&t=1724s) |
| Master loop pattern for agent development | status not stated | [31:40](https://www.youtube.com/watch?v=gd2TycVywv4&t=1900s) |
| BC Code Intel repository | status not stated | [32:04](https://www.youtube.com/watch?v=gd2TycVywv4&t=1924s) |
| Calibrated confidence in agent code | status not stated | [30:33](https://www.youtube.com/watch?v=gd2TycVywv4&t=1833s) |
| Intent steering in agent development | status not stated | [31:30](https://www.youtube.com/watch?v=gd2TycVywv4&t=1890s) |
| Using agents to explain developer commits | status not stated | [34:28](https://www.youtube.com/watch?v=gd2TycVywv4&t=2068s) |
| Managing multiple concurrent agents | status not stated | [35:04](https://www.youtube.com/watch?v=gd2TycVywv4&t=2104s) |
| Agent mental model as overenthusiastic intern | status not stated | [35:28](https://www.youtube.com/watch?v=gd2TycVywv4&t=2128s) |
| Context window limitations and token limits | status not stated | [36:38](https://www.youtube.com/watch?v=gd2TycVywv4&t=2198s) |
| Instructions and custom agent configuration | status not stated | [37:18](https://www.youtube.com/watch?v=gd2TycVywv4&t=2238s) |
| LLM input-output architecture for agents | status not stated | [38:07](https://www.youtube.com/watch?v=gd2TycVywv4&t=2287s) |
| Model selection and cost-performance tradeoff | status not stated | [39:01](https://www.youtube.com/watch?v=gd2TycVywv4&t=2341s) |
| Signal-to-noise ratio in agent context | status not stated | [39:58](https://www.youtube.com/watch?v=gd2TycVywv4&t=2398s) |
| Agent context optimization | status not stated | [39:35](https://www.youtube.com/watch?v=gd2TycVywv4&t=2375s) |
| Agent session reset for fresh state | status not stated | [41:36](https://www.youtube.com/watch?v=gd2TycVywv4&t=2496s) |
| Context window compaction and summarization | status not stated | [41:58](https://www.youtube.com/watch?v=gd2TycVywv4&t=2518s) |
| Specific task decomposition for agents | status not stated | [44:21](https://www.youtube.com/watch?v=gd2TycVywv4&t=2661s) |
| Agentic development as learnable skill | status not stated | [45:33](https://www.youtube.com/watch?v=gd2TycVywv4&t=2733s) |
| Agents on legacy and messy code | status not stated | [46:09](https://www.youtube.com/watch?v=gd2TycVywv4&t=2769s) |
| Documentation assistance via agents | status not stated | [47:58](https://www.youtube.com/watch?v=gd2TycVywv4&t=2878s) |
| Test generation and scenario expansion | status not stated | [48:43](https://www.youtube.com/watch?v=gd2TycVywv4&t=2923s) |
| Scaffolding and table design via agents | status not stated | [49:47](https://www.youtube.com/watch?v=gd2TycVywv4&t=2987s) |
| Design validation and assumption challenging | status not stated | [50:52](https://www.youtube.com/watch?v=gd2TycVywv4&t=3052s) |
| Autonomous agent compilation and testing | status not stated | [51:28](https://www.youtube.com/watch?v=gd2TycVywv4&t=3088s) |
| Agent-driven code improvement through documentation and quality verification | status not stated | [52:20](https://www.youtube.com/watch?v=gd2TycVywv4&t=3140s) |
| XML documentation for code objects improving AI understanding | status not stated | [55:45](https://www.youtube.com/watch?v=gd2TycVywv4&t=3345s) |
| AL guidelines best practices resource | status not stated | [54:20](https://www.youtube.com/watch?v=gd2TycVywv4&t=3260s) |

## AL objects mentioned

As heard in the captions. A name that matches one object page by exact type and name links to it; the others stay as named.

- other "BC Code Intel MCP" at [25:41](https://www.youtube.com/watch?v=gd2TycVywv4&t=1541s)
- other "BC Code Intel" at [40:41](https://www.youtube.com/watch?v=gd2TycVywv4&t=2441s)
- other "MCP servers" at [41:00](https://www.youtube.com/watch?v=gd2TycVywv4&t=2460s)
- other "telemetry buddy" at [41:12](https://www.youtube.com/watch?v=gd2TycVywv4&t=2472s)
- other "object ID Ninja" at [41:12](https://www.youtube.com/watch?v=gd2TycVywv4&t=2472s)

## Quotes

- [7:42](https://www.youtube.com/watch?v=gd2TycVywv4&t=462s) "We bring change to our customers, and now change has come for us."
- [8:08](https://www.youtube.com/watch?v=gd2TycVywv4&t=488s) "you come at things with the I write every line of code. I understand everything that's in here."
- [9:53](https://www.youtube.com/watch?v=gd2TycVywv4&t=593s) "In agentic world, the agent might do different stuff every time. That's a very, very big shift."
- [10:05](https://www.youtube.com/watch?v=gd2TycVywv4&t=605s) "your intent and your ability to sculpt the intent for the agent to understand is now the new mastery."
- [11:58](https://www.youtube.com/watch?v=gd2TycVywv4&t=718s) "matters more than whether or not you're working with ai"

## Disclaimers in the video

- [13:02](https://www.youtube.com/watch?v=gd2TycVywv4&t=782s) subject-to-change: For now, it's optional
- [20:43](https://www.youtube.com/watch?v=gd2TycVywv4&t=1243s) coming-later: there's an upcoming session in June that will be very much around this
- [32:35](https://www.youtube.com/watch?v=gd2TycVywv4&t=1955s) subject-to-change: the agent didn't understand BC well enough to do those things correctly
- [37:05](https://www.youtube.com/watch?v=gd2TycVywv4&t=2225s) subject-to-change: there are other places and models where you can get up to a million, but context window isn't everything
- [51:43](https://www.youtube.com/watch?v=gd2TycVywv4&t=3103s) coming-later: You should be able to get to a point by maybe end of summer where you can try out at ...
- [57:37](https://www.youtube.com/watch?v=gd2TycVywv4&t=3457s) coming-later: one of my projects for this quarter two in the months ahead April through June is looking at what can ...
- [58:08](https://www.youtube.com/watch?v=gd2TycVywv4&t=3488s) coming-later: There may be some updates to AL guidelines coming in the upcoming years upcoming weeks rather to help with providing some additional skills

Presenters (as heard): Luke van Vught, Jeremy Visca, Stefan Møller, Henrik, Luke, Kamel, Vico, Brandon Sanderson, Dean Jackson, Mir Kasim, Peter Konijn.