Agentforce
Readiness Assessment

An Agentforce readiness assessment checks, before you build, whether a use case is worth automating and whether your org can support it. It reviews value and volume, the data and knowledge the agent would ground on, the agent user's access, Trust Layer considerations, the actions you already have and the channels you would use.

Salesforce notes that agents succeed better with up-to-date, accurate and relevant information, and that agent testing can lengthen an implementation. An assessment finds data and access problems while they are still cheap to fix.

Seven areas: the use-case shortlist and its value, data quality and knowledge for grounding, permissions and agent user access, Einstein Trust Layer considerations, actions available from existing Flows, Apex and APIs, channel fit, and the success measures you will judge the agent by. Each area produces findings and a rating.

We sample the knowledge articles, files and records each use case depends on and check them for accuracy, ownership, duplication and gaps. We also confirm whether Data 360 is enabled and which grounding route fits. Content that is out of date or owned by no one goes on the fix list before any build starts.

Trailhead notes that unstructured content such as documents needs preparation before it can be retrieved well. Agentforce Data Library supports Knowledge articles, files and web search; the advanced Data 360 setup joins several sources, including harmonized, zero-copy and real-time data.

We map what the agent would need to read and do, which user it would run as, and which existing Flows, Apex and APIs could become its actions. We then check that plan against the Einstein Trust Layer and your security rules, so sensitive data stays out of reach and every action has an owner.

The rules come from Salesforce. Employee-facing agents use the permissions of the person interacting with them; external-facing agents run as a single dedicated user, and Trailhead advises giving it only the permissions it needs. Flow-based actions must use the Autolaunched Flow (No Trigger) type with at least one input and one output, and a flow's description becomes an instruction for the agent, so names and descriptions are reviewed too.

On the Trust Layer, Trailhead notes that data masking is currently disabled for agents, so we record which controls your security team relies on and what that means for each use case.

You receive a ranked use-case shortlist with baselines, a data and knowledge fix list, an agent user and permissions design, an inventory of reusable actions, a Trust Layer and security notes document, a channel recommendation and a pilot plan for the top use case. The deliverables are yours to use with Techila or another partner.

A business owner for each candidate use case, a Salesforce admin, someone who owns the knowledge base, a developer or architect who knows your Flows, Apex and integrations, and a member of your security or compliance team. A sponsor who can approve the pilot should join the readout.

The pilot plan in the assessment becomes the brief for the Agentforce Pilot: one use case, one channel, a baseline, a test approach and exit criteria. Fixes from the data and permissions lists are made first, so the pilot starts on a ready org rather than discovering problems mid-build.

Scope, duration and price are set in a scoping call, because they depend on how many use cases and teams are involved and how much of the org has to be reviewed. Agentforce licence and usage costs are separate; see the Agentforce pricing guide and the Agentforce cost calculator. To arrange the call, contact Techila.