← All talks
aman sardana. AI/ML/CV Meetup · April 2, 2026

10 lessons from building
AI agents that people
actually use.

Practical patterns from 18 months of production
multi-agent systems in the enterprise.

01 / 15
aman sardana. FIELD NOTES / ABOUT

Hi, I'm Aman.

Aman Sardana

Sr. AI Engineering Architect. 18+ months building multi-agent AI systems over structured enterprise data. Thousands of daily users. Before that: distributed systems, cloud architecture, and a deep belief that the best technology disappears into the workflow.

02 / 15
aman sardana. FIELD NOTES / CONTEXT

Where these lessons come from

18 months building a multi-agent system over enterprise SQL data, used daily by thousands of business users. These 10 lessons shaped whether the system got adopted or ignored.

These will resonate most if you are...

  • Building agents for business users, not analysts or developers
  • Working with structured data (text-to-SQL, database-backed tools)
  • Using multi-agent orchestration (LangGraph, CrewAI, etc.)
  • Trying to move past demos into daily production use
  • Operating with compliance, access control, and internal jargon

The common thread

None of these are about which model or framework to pick. They're about the patterns between the model and the user that determine trust, adoption, and retention.

Some are emerging across the industry (think OpenClaw-style ambient agents). Others are specific to enterprise data. All come from shipping, watching users, and iterating.

Views and opinions expressed in this presentation are my own and not those of my employer. Examples are generalized and do not disclose proprietary information.

03 / 15
aman sardana.
01

Your Users Speak a Language
the Model Doesn't.

Every org has internal shorthand. Acronyms that mean different things in different contexts. LLMs guess confidently. They guess wrong.

Same acronym, different meaning
"What's the NPS trend for our top accounts?"
NPS → Net Promoter Score Customer satisfaction
"Check the NPS configuration on the edge router"
NPS → Network Protection Solutions Networking context

Thousands of internal acronyms per org, many overlapping across departments.

The pattern

Resolve jargon before agents execute. Use surrounding context for disambiguation, not just the term itself.

User query Extract acronyms Context + vector search Inject definitions Agents execute

~50% of user queries benefit from disambiguation.

04 / 15
aman sardana.
02

Assume Sensible Defaults.
Show Your Work.

Clarification questions add friction. LLM accuracy degrades up to 33% in extended multi-turn conversations.
Microsoft Research, ICLR 2026

Friction loop
User: "Show me sales performance"
AI: "Which region?"
User: "North America"
AI: "What time period?"

Three turns for one question. Early wrong turns compound.
Sensible defaults + transparency
User: "Show me sales performance"
AI: All regions, current quarter.
Region: All · Period: Q3 FY26 · Metric: Revenue

Users see the assumptions. They narrow from there.

Data filters transparency

Critical in text-to-SQL. Business users think in portal labels. The DB schema uses different names. "Active" = status='CURRENT'

Show which filters were applied, in language they understand. Lets them catch mismatches. Builds trust.

Intelligent reflection

Results sparse but close match exists? Auto-retry with corrected input. Resolve ambiguity behind the scenes.

05 / 15
aman sardana.
03

Guide Discovery.
Don't Hope for Good Questions.

Most users don't know what your system can do. If the LLM suggests follow-ups, it will hallucinate capabilities you don't have.

Generic follow-up
"Is there anything else you'd like to know?"
Leads nowhere. Conversation dies.
Grounded follow-up
"Would you like me to break this down by product line?"
Specific. Actionable. System can fulfill it. Teaches through example.

~40% of follow-up interactions come from guided suggestions.

The architecture

Decouple from your main agent. Own trigger policies, capability grounding, novelty filtering. Run in parallel, zero added latency.

Trigger policy Capability retriever Generate candidates Novelty filter

Open-sourced this as a dedicated LangGraph node:

06 / 15
aman sardana.
04

Security Is Not an
LLM Problem.

Access control cannot live in a prompt. Not partially. Not with clever instructions. Not at all.

The temptation
"This user can see accounts A, B, C. Only return data from those."

Might work for one query. Might not for the third. Can be jailbroken. You can't tell an auditor access control "usually works."
The only answer
Deterministic. Opaque to the LLM. Model writes queries; code enforces boundaries. When filtering hides data, tell users: "Records exist but aren't visible with your access."
User query LLM generates SQL Code injects RBAC predicates Execute
07 / 15
aman sardana.
05

Empty Results Are Never
"No Data Found."

Three distinct reasons a query returns nothing. Each needs a different response.

Access-filtered

Data exists, user can't see it. Say so: "Records exist but aren't visible with your current access."

Filter mismatch

Entity exists, filters don't match. "No Q3 orders for Acme. Could mean they ordered in Q2."

Query error

Filter values don't exist. Misspelling, wrong ID format. Analyze which filter failed and suggest corrections.

Never let your synthesis layer say "no results" without explaining why.

08 / 15
aman sardana.
06

Personalization Is Not a Feature.
It's the Product.

A system that treats every user the same will plateau on adoption no matter how accurate it is.

Shortcuts

User-defined command aliases. /pipeline west expands to a full multi-line prompt.

5+ shortcuts = 2-3x engagement.

Scheduled delivery

"Every Monday, send me a summary of my open deals." Reactive to proactive. Insights without logging in.

What turns AI from a tool into a habit.

Long-term memory

"This client is price-sensitive." Remembered across sessions. The assistant adapts to you.

Memory feeds suggestion quality. Virtuous cycle.

09 / 15
aman sardana.
07

Your AI Has Blindspots
If It Only Sees Chat.

Users operate in dashboards, spreadsheets, email threads. Their real context lives outside the chat window.

The disconnect
User is viewing Q3 accounts with declining usage. Asks: "What should I prioritize?"
Assistant sees only chat text. Gives a generic answer. Doesn't match what's on screen. Trust drops.
The fix
Pass live UI state (filters, selected objects, data scope) as explicit context to every turn.
Answers grounded in the exact view the user sees. Fewer loops. Feels like an extension of the interface.

The broader principle

The AI should ingest the same stimulus the human sees. Context changes? AI context changes with it.

Embed AI where humans already work. Chat, email, team channels, dashboards. Don't force context-switching.

10 / 15
aman sardana.
08

Let Users Build on Top
of Your Platform.

You can't anticipate every workflow. Give users composable building blocks. Let them assemble their own.

The primitives approach

Expose agents and tools as composable steps. Users chain them in natural language. You provide: gather, decide, act, synthesize. They describe the recipe.

Every step deterministic. No context loss. Full visibility. This takes you from "chatbot" to platform.

User-created workflow
"Every Monday, pull my top 5 accounts, check for declining usage, send me a summary."
System assembles from building blocks, lets them test it, saves it. No code. No tickets.

Distinct from personalization

Lesson #6 = adapting the system to users (shortcuts, memory). This = letting users create new capabilities that didn't exist before.

Best enterprise products become platforms. Users who compose their own workflows become your strongest advocates.

The flow
User describes goal AI assembles steps Preview + test Save + schedule
11 / 15
aman sardana.
09

Meet People Where They Work,
Not Where You Built.

Your users live in email, team channels, and video calls. If AI only works inside your app, it will always have blindspots.

Email delegation

CC the assistant on any thread. It reads context, picks out the ask, runs work in the background. No context-switching.

Team channels

Tag it in a group conversation. Answers questions, pulls data, takes actions alongside the team. Not a 1:1 silo.

Proactive intelligence

Don't wait to be asked. Monitor for changes, detect anomalies, surface insights. Event-driven, not query-driven.

The measure of a great assistant isn't what it can answer. It's how little the user has to remember to ask.

12 / 15
aman sardana.
10

Build the AI System
with AI.

The meta-pattern. The value isn't in general intelligence. It's in encoded constraints that channel intelligence toward useful outputs.

What this looks like

  • Machine-readable rules for code conventions, architecture, domain requirements
  • Structured command files for complex multi-step operations
  • New developers contribute effectively from day one
  • Multi-week sprints collapse to days

The compounding advantage

A model with no context writes generic code. The same model with project conventions writes code that fits your codebase.

Teams that make their codebase legible to AI from the start compound that advantage every sprint. The gap is already a chasm, and it's widening.

13 / 15
aman sardana. FIELD NOTES / TAKEAWAYS

The 10 Lessons

01

Your users speak a language the model doesn't

02

Assume sensible defaults, show your work

03

Guide discovery, don't hope for good questions

04

Security is not an LLM problem

05

Empty results are never "no data found"

06

Personalization is not a feature, it's the product

07

Your AI has blindspots if it only sees chat

08

Let users build on top of your platform

09

Meet people where they work, not where you built

10

Build the AI system with AI

The gap between a technically impressive demo and a product people actually use is filled with these patterns. Some are emerging. Some are foundational. All of them compound.
14 / 15
aman sardana.

Keep building.

More field notes at amansardana.work. Let's connect.

15 / 15