Skip to main content
Dreamforce 2026 / September 15–17 / San Francisco

Tucario at Dreamforce 2026.

Build Trust and Control for Agentforce at Scale.

Join Tucario for a focused 20-minute architecture session on how to give production agents clear boundaries, enforceable controls, and evidence enterprise teams can stand behind.

Date and timeSep 16 / 2:00 PM
WhereMoscone West
RoomRedwood Theater

Until the session

September 16, 2026 at 2:00 PM PDT

0Days
00Hours
00Minutes
00Seconds
Add to my agenda
Real 2025 panorama of downtown San Francisco from Twin Peaks, styled in Tucario's editorial palette
Tucario at Dreamforce 2026
Session09 / 162:00 PM PDT

1 / The session

Ready to ship. You just do not trust it.

Building capability is easy. Confidence is the harder job. A production agent needs explicit boundaries, rules the platform enforces, and evidence your team can defend after the fact.

Trust boundaries

Decide what the agent may access through its mandate, execution context, grounding scope, and least-privilege permissions.

Control mechanisms

Move hard rules out of prompts and into action filters, permissions, Flow, Apex, and deliberate human gates.

Observability layers

Prove the agent before launch, trace every session after launch, and improve it from evidence instead of anecdotes.

Working principle

A prompt is not a security control. Trust is architecture, not a promise.

TRUST CONTROL OBSERVABILITY

2 / The Agentforce Architecture

Three pillars. One architecture teams can operate.

Trust boundaries define reach. Control mechanisms define authority. Observability turns every outcome into evidence. Together they make an agent governable at enterprise scale.

1
What may it access?

Draw the trust boundary.

Write the mandate, choose the execution context, scope grounding, and grant least privilege on purpose.

2
What may it do?

Turn instructions into enforcement.

Use action filters, permissions, Flow, Apex, and escalation so critical rules cannot be talked around.

3
What did it do?

Build an evidence trail.

Combine Testing Center, Session Tracing, and Agent Analytics to test in the lab and monitor in the field.

3 / From soft to hard

The model decides whether. Code decides what.

A prompt can guide behavior, but it cannot guarantee it. Put hard business rules below the prompt line, where action filters, permissions, Flow, Apex, and human judgment can enforce them.

Prompt Guides the model
Platform Enforces the boundary
People Own irreversible decisions
Agentforce robot reading a guide
Artur Kolasa
Michał Bajdek
Sep 16 2026

4 / Dreamforce

Build Trust and Control for Agentforce at Scale.

Meet Tucario at Dreamforce 2026 for a practical conversation about trusted data, clear control, and Agentforce architecture your team can operate.

DreamforceSeptember 15–17, 2026
Time2:00–2:20 PM PDT
VenueMoscone West, L1 Forest
RoomRedwood Theater
SpeakersArtur Kolasa + Michał Bajdek
Save the session

Until the session

September 16, 2026 at 2:00 PM PDT

0Days
00Hours
00Minutes
00Seconds

5 / Dreamforce 2026 FAQ

Plan the session. Bring the hard questions.

Everything you need to find Tucario at Dreamforce and decide whether the session belongs on your agenda.

When and where is Dreamforce 2026?

Dreamforce 2026 takes place September 15–17, 2026, at Moscone Center in San Francisco and online through Salesforce+.

What is Tucario presenting at Dreamforce 2026?

Tucario is presenting “Build Trust and Control for Agentforce at Scale,” a practical architecture session about trust boundaries, enforceable controls, and observability.

When is the Tucario Agentforce session?

The session is on Wednesday, September 16, 2026, from 2:00 to 2:20 PM PDT at Moscone West, Redwood Theater.

Who is speaking?

The session is presented by Artur Kolasa, Salesforce Certified Technical Architect and Tucario co-founder, and Michał Bajdek, Technical Architect and Tucario co-founder.

6 / After the session

Take the architecture back to work on Monday.

The post-session guide turns the talk into a nine-check review your team can run against every production agent.