Deep research / practitioner evidence + academic calibration
What poteto teaches about AI engineering
Lauren Tan’s pstack is best read as a theory of the engineering system around an LLM: make context legible, tools safe, failures observable, and trust measurable.
01 / The system around the model
Why prompt quality is only one layer of the problem
model / inferencegrounding / interfaceverification / memoryhuman judgment
02 / Verification is the scaling primitive
Parallelism only works when the loop can close
03 / Depth before breadth
Why more agents do not automatically make a system more intelligent
04 / Academic calibration
What is supported, what is a hypothesis, and what remains open
Evidence map
Supported
Interface design matters. SWE-agent reports that a purpose-built agent-computer interface improves repository navigation, editing, and test execution. This supports treating tools and affordances as part of the agent’s effective capability.
Supported
Feedback improves loops. ReAct interleaves reasoning and action; Reflexion stores verbal feedback as episodic memory. These mechanisms give academic backing to runtime observation and failure memory.
Supported
Eval the whole system. Anthropic recommends measuring agent performance and OpenAI describes continuous task-level evals for its internal data agent. A transcript is not an evaluation.
Hypothesis
Depth-first orchestration is a default. It is a sensible response to context and verification costs, but it should be measured against simpler single-agent workflows for each workload.
Hypothesis
Hard constraints beat more prose. Constraints can reduce the action space, but rules such as banning a particular React primitive are architecture-specific and require local evidence.
Open
Does the workflow transfer? Coding has unusually strong oracles. General knowledge work, product decisions, and scientific reasoning need richer evaluators and may require more human judgment.
Consulting translation
When advising a team, ask these questions before recommending an agent framework:
- What outcome is being purchased?
Reduce cycle time, improve correctness, widen access, or lower operational cost? - What is the oracle?
Test, database invariant, screenshot, trace, reviewer rubric, or user-confirmed result? - What can the agent touch?
Define identity, authorization, sandboxing, secrets, data boundaries, and reversible actions. - What happens when it is wrong?
Make retryable failures explicit, preserve evidence, and route ambiguous or high-impact cases to a human. - How does one failure improve tomorrow?
Encode it as a regression, tool fix, skill, lint rule, or product constraint.
05 / A practical study route
From LLM literacy to dependable AI systems
Source ledger
Primary or first-party sources used in this synthesis