Forward deployed engineering on Salesforce and everything around it.
An Aquiva forward deployed engineer (FDE) works with you on one specific agent use case. They help you decide whether it's worth building, agree with you which number it should move, build it in your own org, and keep measuring once it's live. Aquiva is a Salesforce Cloud Expert PDO and a Launch Partner of the FDE PDO Partner Network, the specialized PDO category of Salesforce's FDE Partner Network, which Salesforce opened in April 2026. Most of the work is on Agentforce and Data 360. When a use case is better served headless, on Claude, or on a data platform outside Salesforce, we build it there.
FDE PDO Partner Network Launch Partner · Salesforce Cloud Expert PDO · 10 years in business · 70+ verified reviews at 4.91/5
Most of the job happens before and after the build.
Nobody hands an FDE a finished spec. They work out with your business owners what to build, how you'll measure it, and how people will start using it, and then they build it.
Decide what's worth building
Some agent ideas are hard to deliver on Agentforce today, and some can't be measured once they're live. We'd rather tell you that before any sprints are spent on it.
Agree on the number first
Before anyone writes a topic or an action, we agree which metric the agent should move, where it's measured, and what counts as working. Agentforce Observability and Data 360 analytics go in with the agent, which gives you a baseline to compare against after launch.
Build in your org
The first version runs in your org or a sandbox, on your data, permissions, and integrations. It covers less than the full idea, but it's the version you take to production rather than a demo someone rebuilds later.
Stay through rollout
After launch we sort out who gets access, train the first users, and tighten the guardrails when the agent does something nobody expected. Then we report back on whether the metric moved.
Agentforce and Data 360 first, other platforms when the use case needs them.
Salesforce's FDE qualification covers Agentforce and Data 360. Our FDEs hold it, and they also build on the platforms around Salesforce.
Agentforce and Data 360 at the center
Salesforce built its FDE Ready program around these two products, and most of our FDE work is on them. That means agents grounded through retrieval on a vector index, Data 360 doing the integration work, and Zero Copy for data that was never going to be copied into Salesforce. The agent build itself follows our Agentforce practice: trust layer, evals, and versioned prompts.
Headless and MCP
People use an agent where they already work, and more and more often that isn't a Lightning page. We build MCP servers that expose Salesforce data and actions to agents, and we put the agent in Slack, Claude, Microsoft Teams, or a custom UI your own team owns. For one customer we stood up an MCP server over a CPQ sandbox, so a rep could build a full quote, product lines and pricing rules included, from a single prompt. Salesforce is moving the same way with the Headless Experience Layer it showed at Dreamforce 2026, and our Headless Salesforce POV covers the patterns.
Anthropic and Claude
Aquiva is a member of Anthropic's Claude Partner Network. We build agents on Claude through the API and the Claude Agent SDK, connect Claude to Salesforce and the rest of your systems through MCP, and help engineering teams adopt Claude Code with the costs and the review process under control. We use Claude on our own delivery too: a pipeline that takes a GitHub ticket to a reviewed pull request, written up here, and Claude Code agents doing manual QA on Salesforce projects. AQUA, the assistant on this site, runs on Claude as well.
When the data lives outside Salesforce
Some customers already run their data and AI on another platform: Bedrock on AWS, Microsoft Foundry on Azure, Gemini Enterprise Agent Platform (formerly Vertex AI) on Google Cloud, Cortex AI on Snowflake, Agent Bricks on Databricks. Our FDEs build there when that's where the use case belongs. The data-layer work runs through Data & Integrations.
Three phases, and you decide after each one.
We start with one use case. Each phase ends with a readout, and you decide from it whether to go on.
Validate
A short engagement on one or two use cases. We qualify them, set the KPIs and the baseline, deploy a first version into your environment, and gather feedback from the people who'll use it. It ends with a recommendation to scale it, change it, or stop.
- Use-case qualification
- KPIs and baseline
- First deployment in your org
- Scale, change, or stop readout
Build to production
The validated use case goes to production with an FDE leading, alone or with a small team. For a small, nimble organization that's often about a month; an enterprise solving something bigger should plan on a couple.
- Production build and integrations
- Guardrails and evals
- Rollout and user enablement
- KPI readout against the baseline
Operate
Agents drift as the data, the prompts, and the platform change under them. A lighter monthly engagement keeps measuring, fixes what isn't performing, and adds use cases and users as you go. It can sit inside Salesforce Managed Services.
- Monthly KPI reporting
- Tuning and re-architecture
- New use cases
- Enablement for new teams
Forward deployed means working inside your team: in your Slack, your standups, and the huddle when something breaks.
We’ve worked inside customer teams for 10 years.
Our engineers have always worked in customers' Slack channels and daily huddles, and the person who built a feature is usually on the call about it. What's new is the agentic skill set. Our FDEs go through Salesforce's FDE Ready program, which requires the Agentforce Specialist and Data 360 Consultant certifications, Agentblazer Legend status, and the implementation readiness courses for both products before the workshop itself. We put consultants through it as well as engineers, because the role depends more on how someone works with you than on their job title. It also means the person who scopes the use case reads the KPIs the same way as the person who builds it. Robert Sösemann, our Director of Engineering, wrote up what the program covers after the Munich workshop.
How we work publishes the delivery numbers behind this: 87% of our pilots reach production, and getting there takes under 90 days. From AI pilot to production →
Common questions
What is a forward deployed engineer?
Someone who works inside the customer's team to get an agent into production and prove it works there. On Salesforce that means qualifying the use case, agreeing the KPIs before the build, building on Agentforce and Data 360 in the customer's own environment, and staying through adoption. The title comes from Palantir. Salesforce runs FDE teams of its own and opened the model to partners through the FDE Partner Network in April 2026.
Is an FDE an engineer or a consultant?
The title covers two profiles, and most engagements need both. One writes production code inside your environment and owns the integrations. The other, often called a deployment strategist, runs the pilot with your business owners: which use case, which KPI, and how the team adopts it. Salesforce's own FDE teams pair one deployment strategist with two FDEs. We staff FDE work from both our consulting and engineering teams.
How is an Aquiva FDE different from a Salesforce FDE?
The qualification is the same: our FDEs go through FDE Ready, and Aquiva is a Launch Partner of the FDE PDO Partner Network, the specialized PDO category within Salesforce's FDE Partner Network. The difference is what they build on. Alongside Agentforce and Data 360, we build headless agents behind MCP servers, agents on Claude as a member of Anthropic's Claude Partner Network, and data work on platforms like Snowflake and Databricks.
Does forward deployed mean on-site?
No. Most of the work runs remotely, and we travel when a workshop or a rollout needs people in the room.
How long does an FDE engagement take?
Validate is a short engagement on one or two use cases that ends in a scale, change, or stop recommendation. Build to production is often about a month for a small, nimble organization and a couple of months for a larger enterprise problem. Operate runs monthly after that.
What happens after the agent goes live?
The Operate phase keeps measuring the KPIs agreed at the start, reworks what isn't performing, adds use cases, and enables new teams. It can fold into Salesforce Managed Services, which puts the agent under SLA-backed support.
Bring the use case you’re not sure about.
An FDE engagement starts with a 30-minute call about one use case and how you'd know it worked.