How Senior Engineers Actually Build with AI in 2026 | Build a Splitwise AI Clone
This video argues the junior-vs-senior engineering gap hasn't closed with AI, it has moved to system orchestration, then walks through building 'Splitter AI' (a Splitwise clone with AI-parsed expenses and debt-simplification) using a markdown-only context/skills architecture (agents.md, a skills folder, and a 33-item Next.js checklist) plus agent-native tools like the Clerk CLI and Prisma's official AI skills pack.
Ankita Kulkarni100 minTranscript found
Quick learning frame
Read this before watching.
A model becomes useful when it is wrapped in a harness: tools, state, permissions, memory, routing, and verification.
New playlist item from Ankita Kulkarni; queued for transcript-backed review, topic mapping, and a practical learning artifact.
Skill you build: The ability to architect an agent workflow around durable markdown context (project docs, skills, checklists) and agent-native CLIs instead of letting the AI freehand custom infrastructure or requiring manual dashboard clicks.
Watch for the shift from claim to mechanism. The learning value is the point where the transcript reveals a repeatable action, tool boundary, context move, review habit, or artifact.
Concept diagram
Where this video fits.
01User intent
02Model role
03Tool surface
04State and memory
05Verification loop
06Reusable operating rule
Deep lesson
Turn this video into working knowledge.
17,173 cleaned transcript words reviewed across 4,718 timed caption segments.
Thesis
How Senior Engineers Actually Build with AI in 2026 | Build a Splitwise AI Clone teaches a practical agent harness move: This video argues the junior-vs-senior engineering gap hasn't closed with AI, it has moved to system orchestration, then walks through building 'Splitter AI' (a Splitwise clone with AI-parsed expenses and debt-simplification) using a markdown-only context/skills architecture (agents.md, a skills folder, and a 33-item Next.js checklist) plus agent-native tools like the Clerk CLI and Prisma's official AI skills pack.
The goal is not to remember the video. The goal is to extract the operating principle, tie it to timestamped evidence, test how far the claim transfers, and make something reusable.
0:19
Orchestration over syntax
“Here's what's actually happening. If you ask an AI agent to build a full stack application today, it will happily write 2,000 lines of custom session management. bast JWT logic and complex database triggers from scratch. A junior...”
The creator argues a senior engineer's job in 2026 is no longer writing syntax but orchestrating systems: an AI agent will happily write 2,000 lines of custom session management and JWT logic from scratch, which a junior sees as a superpower but a senior recognizes as an undocumented maintenance nightmare, so seniors provision industry-standard tools instead of letting AI reinvent infrastructure. Next time an agent proposes custom auth, session, or database-trigger code, stop and ask whether an established provider (like Clerk or Prisma) already solves it before accepting the generated code.
30:16
Skills as living checklists
“that's what you want to use. So on top of that, we are going to add more skills in the future when we add to Prisma. So the full stack for this whole video is context is what...”
Each skill is just a markdown file with a description (read at the start of every session so it auto-fires when the task matches) and a body payload, like the Next.js review skill's checklist that grew from 29 to 33 items as new bugs (like a button silently not submitting) were discovered and appended, so the agent effectively maintains its own training material over time. Create one skill file for a recurring mistake your AI agent makes, and add a rule to it the next time that mistake happens instead of just fixing the code once.
77:24
Logic before polish
“huge believer of that you don't need to use the most expensive models for everything. Just pick and choose. For example, if you have a budget or an application like this where you need some AI capabilities, but...”
The creator deliberately builds application logic (auth, AI quick-add parsing, debt calculations) before touching UI polish, then moves to design system and theming with Shadcn UI at the end, saying it's fine to sequence it either way but he personally prefers logic first so the AI has stable context before restyling. On your next build, decide upfront and write down whether you'll sequence logic-first or UI-first, then stick to it so your agent's context stays consistent.
01
User intent
Start with this video's job: This video argues the junior-vs-senior engineering gap hasn't closed with AI, it has moved to system orchestration, then walks through building 'Splitter AI' (a Splitwise clone with AI-parsed expenses and debt-simplification) using a markdown-only context/skills architecture (agents.md, a skills folder, and a 33-item Next.js checklist) plus agent-native tools like the Clerk CLI and Prisma's official AI skills pack. Treat "User intent" as the outcome you are trying to make visible, not a topic label. Anchor it to 0:19, where the video says: “Here's what's actually happening. If you ask an AI agent to build a full stack application today, it will happily write 2,000 lines of custom session management. bast JWT logic and complex database triggers from scratch. A junior...”
02
Model role
Use "Model role" to locate the part of the agent harness mechanism the video is demonstrating. Ask what changes in your real setup if this claim is true. Anchor it to 30:16, where the video says: “that's what you want to use. So on top of that, we are going to add more skills in the future when we add to Prisma. So the full stack for this whole video is context is what...”
03
Tool surface
Turn "Tool surface" into the reusable artifact for this lesson: A one-page agent harness map with tool boundaries, state ownership, and proof signals. This is where watching becomes something you can inspect and reuse.
04
State and memory
Use "State and memory" as the application surface. Decide whether the idea touches a browser flow, a local file, a model choice, a source document, a UI, or a review step.
05
Verification loop
Use "Verification loop" to prove the lesson. The evidence should connect back to the video title, transcript anchors, and a concrete output, not a generic best-practice claim.
06
Reusable operating rule
Use "Reusable operating rule" to carry the idea forward: save the prompt, checklist, diagram, or operating rule that would make the next agent run better.
Example
Source-backed artifact packet
Convert the video into a scoped artifact request that includes the transcript claim, mechanism, acceptance criteria, and proof. The output should be a one-page agent harness map with tool boundaries, state ownership, and proof signals..
Example
Agent harness proof brief
Separate what the speaker claims, what the demo actually proves, and what still needs outside verification before you adopt the agent harness pattern.
Example
Teach-back module
Transform the lesson into a definition, a User intent -> Model role -> Tool surface -> State and memory -> Verification loop -> Reusable operating rule diagram, one misconception, one practice exercise, and a check-for-understanding question.
Do not learn it wrong
Treating the title as the lesson without checking what the transcript actually says.
treating model choice as architecture
ignoring tool permissions
missing verification evidence
Letting the lesson drift into generic agent definitions.
Letting the lesson drift into model leaderboard claims.
Letting the lesson drift into tool list without operating boundaries.
Do not count this as learned until these are true.
01
State the transcript-backed claim in your own words: This video argues the junior-vs-senior engineering gap hasn't closed with AI, it has moved to system orchestration, then walks through building 'Splitter AI' (a Splitwise clone with AI-parsed expenses and debt-simplification) using a markdown-only context/skills architecture (agents.md, a skills folder, and a 33-item Next.js checklist) plus agent-native tools like the Clerk CLI and Prisma's official AI skills pack.
02
Explain the practical stakes without hype: New playlist item from Ankita Kulkarni; queued for transcript-backed review, topic mapping, and a practical learning artifact.
03
Map the idea onto the User intent -> Model role -> Tool surface -> State and memory -> Verification loop -> Reusable operating rule sequence and name the weakest link.
04
Produce the artifact and include the evidence that proves it: A one-page agent harness map with tool boundaries, state ownership, and proof signals.
Put it into practice
Give this grounded prompt to Codex or Claude after watching.
You are helping me turn one specific YouTube video into real, durable learning.
Source video:
- Title: How Senior Engineers Actually Build with AI in 2026 | Build a Splitwise AI Clone
- URL: https://www.youtube.com/watch?v=uqX2hY7CseU
- Topic: Interfaces + Open Design
- My current learning frame: Set up a small project with a context folder (project overview, architecture, build plan, schema) plus one custom skill file with a growing mistake checklist, and have an agent build one feature end-to-end using only that markdown context.
- Why this matters: New playlist item from Ankita Kulkarni; queued for transcript-backed review, topic mapping, and a practical learning artifact.
Transcript anchors from this exact video:
- 0:19 / Evidence 1: "Here's what's actually happening. If you ask an AI agent to build a full stack application today, it will happily write 2,000 lines of custom session management. bast JWT logic and complex database triggers from scratch. A junior..."
- 3:40 / Evidence 2: "created to catch app order mistakes before every commit and a strict landing page design spec to keep our UI pixel perfect. This is how senior engineers build. We design solid systems and we orchestrate the AI to..."
- 10:22 / Evidence 3: "that is a starter repo that will have some context folders and some files that will really help you get started with it. And this is how the starter repo looks like. So, before let the agent write..."
- 15:18 / Evidence 4: "checkboxes. The agent always knows exactly what's next and what's already done even in a brand new session. For example, this is how we are going to build it. Now, in this specific file, it's one phase, one..."
- 17:48 / Evidence 5: "have some custom skills that I will walk you through as well. But agents MB file tells the agent to read this folder before writing any code. The docs hold the what and the why. The prompts just..."
- 30:16 / Evidence 6: "that's what you want to use. So on top of that, we are going to add more skills in the future when we add to Prisma. So the full stack for this whole video is context is what..."
- 77:24 / Evidence 7: "huge believer of that you don't need to use the most expensive models for everything. Just pick and choose. For example, if you have a budget or an application like this where you need some AI capabilities, but..."
Video-aware target:
- Prompt lane: Agent harness
- Mechanism to extract: Identify what surrounding harness makes the model more useful than chat alone.
- Artifact to produce: A one-page agent harness map with tool boundaries, state ownership, and proof signals.
- Artifact must include: model role; tools; state/memory; permission boundary; verification proof
Your task:
1. Use the transcript anchors above as the primary source packet. If you add outside context, label it clearly as outside context and keep it secondary.
2. Create a source-check table with columns: timestamp, claim, transcript support, what the demo proves, confidence, and what still needs verification.
3. Extract the actual teachable mechanism from the video: Identify what surrounding harness makes the model more useful than chat alone. Do not invent claims that are not supported by the title, lesson frame, or transcript anchors.
4. Build a reusable learning artifact: A one-page agent harness map with tool boundaries, state ownership, and proof signals.
5. Include:
- a plain-English definition of the core idea
- a diagram or structured model using this sequence: User intent -> Model role -> Tool surface -> State and memory -> Verification loop -> Reusable operating rule
- answers to these source questions: What does the video claim the agent can do? | What surrounding system makes that claim plausible? | What proof is shown instead of merely asserted?
- 3 concrete examples that apply the video idea to real agentic work, such as a repo-editing harness; a local research assistant; a recurring refresh agent
- 2 failure modes the video helps prevent, chosen from the transcript evidence and these likely risks: treating model choice as architecture; ignoring tool permissions; missing verification evidence
- a checklist for the next real workflow, focused on: tool boundaries, state ownership, done signal, recovery path
- one practical exercise with a clear done signal: Map one current coding workflow as a harness and mark the first missing proof signal.
6. Add a "learning transfer" section: what changes in my workflow tomorrow if I actually learned this?
7. Add a "source check" section that cites which transcript anchor supports each major takeaway.
Quality bar:
- Make this specific to "How Senior Engineers Actually Build with AI in 2026 | Build a Splitwise AI Clone", not a generic Interfaces + Open Design essay.
- Tie each harness element to a transcript anchor that names a tool, state boundary, permission, model behavior, or verification step.
- Prefer operational examples, failure modes, and reusable artifacts over broad definitions.
- Call out uncertainty instead of smoothing over weak evidence.
- Avoid these generic drifts: generic agent definitions; model leaderboard claims; tool list without operating boundaries.
- If evidence is weak or missing, stop and say what transcript segment or timestamp needs review instead of guessing.
- Finish with a concise artifact I could paste into my learning app.
Misconceptions
What to stop believing.
A beautiful page is automatically a good learning tool.
Learning requires sequence, active recall, feedback, and application.
Generated UI should be accepted as-is.
Generated UI needs critique, revision, and browser verification.
Practice studio
Learning only counts when you make something.
01
Transcript evidence map
Separate what the video actually says from what you already believe about the topic.
3 source-backed takeaways with timestamps, confidence, and a transfer note.02
One useful artifact
Apply the video to a real workflow and produce a one-page agent harness map with tool boundaries, state ownership, and proof signals..
A reusable artifact with a done signal and one verification step.03
Agent harness teach-back card
Explain the agent harness mechanism to someone who has not watched the video yet.
A 90-second explanation, one diagram, one example, and one misconception to avoid.
Recall check
Answer first, then reveal — without rewatching.
Why does the video say a junior developer and a senior developer react differently to an AI agent that writes 2,000 lines of custom session management from scratch?
How did the Next.js review skill's checklist grow from 29 to 33 items?
In what order does the creator prefer to build a feature, and what does he do at the design-system/theming stage near the end?
Source shelf
Use the video as a doorway, then verify with primary sources.
