Duplex and CopilotKit CLiF,
side by side.
CopilotKit CLiF turns real user workflows into reviewed skills, evaluations and training data. Duplex learns from every result while the work happens, and acts on what it learned, within your rules.
Comparison current as of 16 October 2026.
Every capability: Duplex and CopilotKit CLiF.
What each one does, group by group, from what it is to what it costs.
- Native
- Built into how the product works at its core.
- External
- You provide it, or assemble it, separately.
- Not native
- The product’s documentation doesn’t list it as a built-in feature. A custom build may still be possible.
| Capability | Duplex | CopilotKit CLiF |
|---|---|---|
| What each one is | ||
| What it is | A runtime intelligence beside your AI: it learns and acts for your agents and assistants, within your rules | A full-stack pipeline that turns trajectories into skills, evaluations and training data |
| What it carries forward | Context, decisions, outcomes, rules, tasks, skills and what it has learned | Threads, trajectories, insights, skills and evaluations |
| What gets better with experience | Its next decisions: confidence, routing, recommendations, tool use and reusable knowledge | The agent’s instructions, through reviewed skills drawn from real workflows |
| Knowing what happened, and learning from it | NativeIt knows what happened, and learns what to change next | Learning is generated from stored trajectories |
| Learning | ||
| Continual learning without retraining | NativeIt keeps learning, outside your model’s weights | NativeSkill learning; fine-tuning data is optional |
| Turn-by-turn adaptation | NativeEach new result can change the next decision at once | Analysis runs by hand or on a schedule |
| Learning from corrections | NativeA correction changes what Duplex does next | User changes and app actions feed its learning |
| Learning from outcomes | NativeSuccess and failure feed straight back into what it learns | Outcomes across the stack become insights, skills, evaluations and training data |
| Learning from rewards | NativeA reward strengthens what worked and corrects what didn’t | Can produce fine-tuning data from evaluated trajectories |
| Credit for the steps behind a result | NativeIt works out which earlier actions led to a later result | Analysing trajectories can surface useful workflow patterns |
| How actions move the work | NativeIt learns how each action moves the work from one state to the next | Can turn workflows into skills |
| Predicting what an action will do | NativeFrom what it observes, it predicts what an action will do, tries alternatives and plans | ExternalLeft to the agent framework |
| Better choices of action | NativeOutcomes improve which action it picks next | Reviewed skills improve the agent’s next instructions |
| Reusable skills | NativeExperience becomes skills that are tested before use, and can be paused, brought back or withdrawn | NativeTrajectories become reviewed skills, backed by evidence |
| Testing and managing skills | A skill is tested before it is relied on, and can be paused, brought back, withdrawn or removed for good | Published after human review, delivered at a pinned version |
| Generalising to related cases | It carries what it learned to related cases and new wording, and into skills it can reuse | Skills capture patterns across many production trajectories |
| Deciding and acting | ||
| Knowing how sure it is | NativeBefore every decision, it knows how familiar the case is and how sure it is | Driven by evaluation, not a live measure of certainty |
| Answer, hand over or decline | NativeDuplex answers itself, calls your model, uses a tool or declines, within your rules | ExternalThe host agent decides how to act |
| Answering familiar work itself | NativeIt makes familiar decisions itself, without another frontier-model call | The model still reads the learned skill |
| Handing over what is new or uncertain | New, uncertain or sensitive work goes to the right model, tool or person | ExternalThe host agent’s job |
| Planning, simulation and recommendations | NativeIt predicts what happens next, tries alternatives, builds plans and recommends actions | ExternalThe host agent’s job |
| Many agents, working together | It sends work to the right agent, splits it within your rules, keeps each agent to its scope and brings the results together | Works across supported agent frameworks |
| Learning which tools work | It learns which tools work, when to use them and what each one leads to | Full-stack trajectories capture tool and user actions for new skills |
| Tool actions within your rules | Tools run only through your rules and permissions | Controlled by the host agent framework |
| Streaming | It streams model and tool output, with every step still within your rules | Available through supported agent frameworks |
| Work that carries on | ||
| Learning from every step, automatically | Every step, through MCP + hooks or the plugins, feeds one shared learning loop | AG-UI captures agent and user actions as full-stack trajectories |
| Context across sessions | NativeWork, decisions, tasks and the context that matters carry on across sessions | Durable threads keep app interactions |
| Long-running work | NativeLong work stays open across sessions and agents, with safe points to go back to, until it is finished or forgotten | Durable threads keep the history of interactions |
| Across models | NativeWhat it learns and your context stay outside the model, and follow the work across providers | Strong: skills can be delivered across supported agent stacks |
| Across apps and agents | The same work carries on across Claude, Codex, Cursor, Hermes and your own agents | Skills can be delivered through supported framework adapters |
| Handing work to another agent | A task and its context pass to another agent you allow, ready to continue | Shared trajectories and published skills keep behaviour consistent |
| Keeping context compact | Context stays compact, focused on the rules, tasks and decisions that matter | Trajectories are distilled into insights and skills |
| Learning from images, audio, video and PDFs | It learns from images, audio, video and PDFs, as it does from text | Depends on the captured app events and the agent stack |
| Across media | A question in words can draw on what was learned from images, audio and video, and media can shape later decisions | Outside CLiF’s main focus |
| People, teams and sharing | ||
| For each person | NativeEach person gets their own private context, preferences and learning | Possible through app threads and learning containers |
| For each team | NativeTeams share approved learning, and private work stays private | Published skills spread team learning |
| Across the organisation | NativeWhat works can move from one person to the team and the organisation, under your rules | Reviewed skills can be spread across supported agents |
| Sharing between people, under control | You choose what is shared and with whom, and can withdraw it; anything shared shows who it came from | Project and deployment permissions |
| Protection for shared content | What someone shares arrives as information, never as instructions, so it can’t steer another person’s agent | Developer policy and tool restrictions outrank skills |
| Control and trust | ||
| Your rules | Your rules are part of how Duplex works: they allow, decline or limit real actions | Learned skills can’t override developer safety rules or tool restrictions |
| Rules applied before each action | NativeThe hooks check each action against your rules before it runs | Stays with the host app and the agent framework |
| People in the loop | A person confirms the actions you choose, and what Duplex learns can be reviewed and undone | A strong review before learned skills are published |
| A history you can check | A history of every decision, where it came from and what came of it, that can’t be quietly changed | Insights backed by evidence, and their source trajectories |
| Receipts you can verify | Decisions, skills and deletions can come with receipts you can verify | Links to evidence and exact skill revisions |
| Undoing what was learned, exactly | NativeOne learned change can be undone exactly, without retraining anything | Published skills can be revised or withdrawn |
| Forgetting that reaches everything built on it | Removing something also removes what came from it: context, anything shared from it and related learning | Means removing the affected insights, skills and derived assets |
| Keeping customers apart | Every user, team, organisation and customer is kept apart, in every integration and in everything it shares | Isolation per project and deployment |
| Where everything learned came from | For everything it learned, you can see its source, its scope and how far it can be trusted | Insights and skills stay linked to the trajectories they came from |
| Where it runs and connects | ||
| Any model provider | One intelligence across OpenAI, Anthropic, Google, AWS and compatible models | Works across supported agent frameworks and models |
| MCP | NativeThrough MCP + hooks, Duplex joins every step your agents take | Skills can reach agents and MCP servers through integrations |
| Hooks and plugins | NativePlugins for Claude, Codex, Cursor and Hermes, and hooks that bring Duplex in at each step and check sensitive tool actions | AG-UI and framework adapters |
| Your cloud, or self-hosted | Vareon Cloud, your cloud or no cloud: private cloud, your own account, self-hosted or air-gapped | Cloud-hosted and self-hosted intelligence |
| No cloud: air-gapped | Built for deployments where nothing may leave, with local models, local dependencies and your own identity provider | Possible with a self-hosted stack and local dependencies |
| Cost, speed and fit | ||
| How it lowers model cost | Duplex answers familiar work itself, with no repeated frontier-model call | Better skills cut failures and rework, but the model still runs |
| How it lowers latency | Decisions it is sure of come back fast; uncertain work still goes to your frontier model | Skills make execution more direct |
| Survives a model upgrade | YesLearning, context, tasks and rules stay outside any one model | YesSkills and trajectories stay outside the model’s weights |
| Fine-tuning needed | NoFine-tuning becomes an optional step up | NoNot for skills; CLiF can also produce fine-tuning data |
| Best for | Organisations that need their AI agents to learn, act within rules, carry work forward and improve across models, as one system | Product teams turning real user workflows into reviewed skills, evaluations and training assets |
The other comparisons.
- Duplex and LettaStateful agents with memory and skills
- Duplex and Deeplake HivemindShared memory for coding agents
- Duplex and memory platformsMem0, Zep, AgentCore Memory, Supermemory
- Duplex and fine-tuningFine-tuning and reinforcement fine-tuning
- Duplex and AI gatewaysPortkey, LiteLLM and similar
- The full comparisonDuplex beside all six, side by side
See Duplex on your own work.
Compare it where it counts: a pilot puts Duplex on your own work, with the measures agreed before it begins.