Step 01
Open a Thread
Create a Thread inside any Space. The Thread inherits the Space's knowledge, configuration, and access controls from the start.
Threads
Work with agents the way you actually think.
Threads are real-time conversations where people and agents work together inside a Space. Ask questions, direct work, bring in colleagues, and reach answers together — all in one live conversation.
Thread · launch messaginglive
youtagged in order
@research @copywriter draft the launch messaging for Tuesday.
@researchresponded 1st
Pulled the last three launches and the competitor positioning from Space knowledge.
@copywriterresponded 2nd
Three headline options drafted on top of that research. Ready for review.
@brand leadjoined the thread
Option two. Tighten the subhead and we ship it.
agents working
The gap between asking and doing
Most AI tools answer questions. Threads get things done.
A chat window that gives you a response is useful. A conversation where you can direct multiple agents, control the sequence, involve your colleagues, and refine in real time until the work is finished is something entirely different.
That is what Threads are built for.
01
An answer is not an outcome
A chat window that returns a response is useful. But the work rarely ends at the response — it ends when something is finished, reviewed, and agreed on.
02
One assistant is not a team
Real work draws on research, drafting, analysis, and review. A single conversation with a single model cannot bring those capabilities together in sequence.
03
Colleagues end up somewhere else
When the people who need to weigh in live in a different tool, the context gets copied, pasted, and lost on the way there.
Threads
A shared workspace for people and agents.
A Thread is a conversation created inside a Space. It is where real-time, collaborative work happens — between people and agents, together.
- 01
Add people and agents
Start by adding whoever belongs in the conversation. People and agents both participate in a Thread once added. Collaboration is intentional, not accidental.
- 02
Invoke agents with @
Tag an agent with @ to ask it to respond. Agents already added to the Thread are ready to contribute the moment they are called on.
- 03
Control the sequence
Tag multiple agents in a single message and they respond in the order they were tagged. The flow of a multi-agent conversation is predictable and entirely in your hands.
- 04
Bring colleagues in
Tag a colleague to send them a notification and pull them into the conversation. People and agents work in the same Thread — not in separate silos.
- 05
Grounded responses
Every agent responds with the Space's shared knowledge and its own Skills, Connectors, Tools, Widgets, and Guardrails behind it. Answers are informed, capable, and context-aware.
- 06
One shared conversation
Everyone works from the same Thread and the same agent responses. No duplication, no version drift, no separate conversations producing different results.
Thread · Q3 launch brief5 participants
You@research@analyst@writerBrand lead
Youtagged 3 agents
@research @analyst @writer put together the Q3 launch brief.
q3-goals.pdf4 pages
@researchstep 01
Gathered prior launches and competitor positioning from Space knowledge.
market-scan.md12 sources
competitors.csv38 rows
@analyststep 02
Sized the segments and flagged the two with the strongest pipeline.
segment-model.xlsx3 tabs
@writerstep 03
Brief drafted on top of both. Ready for a human check.
q3-launch-brief.docxdraft v1
Brand leadnotified · joined
Approved with one edit to the positioning line. Ship it.
q3-launch-brief.docxapproved v2
outcome reached · one shared thread
How Threads work
From question to outcome in a single conversation.
Step 02
Add your participants
Add the agents and people who need to be part of this conversation. Participation is intentional — everyone added is ready to contribute.
Step 03
Direct the work
Tag agents with @ to invoke them. Tag multiple agents in one message and they respond in order. Ask follow-up questions, redirect, refine — the conversation evolves in real time.
Step 04
Loop in colleagues
Tag a colleague with @ and they receive a notification. They join the same Thread — reviewing agent responses, adding input, and collaborating alongside the agents in one shared conversation.
Step 05
Reach the outcome
Keep going until the work is done. Adjust direction as the conversation unfolds, without needing to set everything up in advance.
Why Threads
Real-time collaboration, purpose-built for agent work.
Work through problems in real time
Direct agents as the conversation unfolds and adjust course immediately. No need to specify everything upfront before the work begins.
Pull the right expertise together
Add multiple agents to a single conversation so different capabilities — research, drafting, analysis, review — come together around the same problem.
Keep the right people in the loop
Tag colleagues to notify them and bring them into the live conversation. People and agents collaborate together in one Thread rather than in separate channels.
Control who responds and when
Tag order is response order. Multi-agent conversations are predictable, sequenced, and intentional — not chaotic or unpredictable.
One conversation, no duplication
A shared Thread means everyone works from the same set of responses. No duplicate conversations, no conflicting outputs, no wasted effort.
Threads in action
How teams put Threads to work.
Three live conversations, each sequencing agents and colleagues toward a finished outcome.
Case 01
Marketing
Product launch messaging
A marketing manager preparing a product launch opens a Thread in the team's Space and adds a Research Agent, a Copywriter Agent, and two colleagues. In a single message she tags the Research Agent first and the Copywriter Agent second — asking for a competitor positioning summary followed by launch messaging built on it. The Research Agent responds first; the Copywriter Agent builds directly on those findings. She then tags both colleagues, who join the same Thread, review the responses, and refine the tone in real time.
- Participants
- 2 agents · 2 colleagues
- Sequence
- Research → copy
- Outcome
- Messaging refined in-thread
Case 02
Consulting
Client proposal preparation
A consulting team lead needs a proposal response before an afternoon deadline. She opens a Thread in the client Space and adds an Analysis Agent and a Writing Agent, along with a senior colleague to review. She tags the Analysis Agent for background on the client's industry and prior work, then tags the Writing Agent to draft the proposal structure from that context. With both outputs in the same Thread, she tags her colleague for a quick review — and the proposal is ready ahead of schedule.
- Participants
- 2 agents · 1 reviewer
- Sequence
- Analysis → draft → review
- Outcome
- Proposal ready ahead of deadline
Case 03
Support
Technical escalation triage
A support team lead receives a complex escalation and is not sure which team should own it. He opens a Thread and adds a Knowledge Agent configured with product documentation, a Process Agent familiar with escalation routing, and an engineering SME. He describes the issue and tags the Knowledge Agent first, then the Process Agent. The Knowledge Agent identifies the relevant product behavior; the Process Agent recommends a routing path. He tags the SME with the Thread context already in place, who confirms the routing — and the escalation is resolved without a separate chain of emails.
- Participants
- 2 agents · 1 SME
- Sequence
- Knowledge → routing → confirm
- Outcome
- Resolved without an email chain
Secure by design
Every conversation is governed and controlled.
01
Role-Based Access Control
Thread access is governed by Space-level permissions. Only members with appropriate roles can create Threads, add participants, or view conversation history.
02
Audit Log
Every Thread interaction is logged, giving administrators full visibility into who participated, when, and what was produced.
03
Guardrails
Every agent in a Thread operates within its configured Guardrails. Boundaries are enforced at the agent level, not added on afterward.
04
Zero Data Retention
Opal operates on a Zero Data Retention basis with AI model providers. Conversation content is not used to train external models.
05
Certifications
ISO 27001, ISO 42001, and SOC 2 Type II certifications are underway.
Threads across Opal
Every conversation is connected to everything your team needs.
01SpacesThreads are created inside a Space, inheriting its shared knowledge, configuration, and access controls.02AgentsAgents are added to a Thread and invoked with @ to contribute their capabilities to the conversation.03KnowledgeEvery agent in a Thread draws on the Space's Knowledge to ground its responses.04SkillsAgents apply their Skills inside a Thread when invoked, bringing specialized capabilities into the conversation.05ConnectorsAgents use Connectors to pull in external data and surface it directly inside the Thread.06GuardrailsEach agent's Guardrails apply within a Thread, keeping responses within defined boundaries.07TasksTasks run work automatically and asynchronously; Threads are for real-time, directed conversations.08FlowsFlows coordinate structured, multi-step processes; Threads are open-ended, live conversations.
Threads
Bring people and agents together in one conversation.
Threads make real-time collaboration with agents as natural as a conversation. Start building on Opal today.