A Freshdesk ticket carries a customer, a status, and an SLA clock. It does not carry the agent's reasoning about why this ticket is at risk, which policy applies, or which support-ops lead signed off on the recovery plan. When Freddy AI drafts a response or escalation, the trail of that decision lives in a chat panel that nobody reads next quarter. Dock keeps the brief, the approver, and the pointer back to the Freshdesk ticket in one row that an auditor can open.
Freshdesk and Freddy AI stay the system of record for the raw data. Dock is the system of record for what the AGENT INTERPRETS. Each Dock row carries a pointer back to the platform record, agent identity, decision, reviewer, and timestamp. The agent re-fetches platform data via fresh API reads when it needs current state.
The SLA Brief table
One Dock table, owned by the support-ops lead. Each row is one at-risk ticket the agent flagged with a recommended path.
| Freshdesk ticket | Risk read | Recommended path | Drafting agent | Support-ops approver | Decided at |
|---|---|---|---|---|---|
| FD-48201 (Tier-2, 4h left) | Customer replied at 02:14 UTC, response window tight | Reassign to APAC queue, send holding reply | freddy-triage@dock | Priya Anand | 2026-05-30 09:12 |
| FD-48244 (Tier-1, breached 22m) | Bug report needs eng confirmation before refund | Open linked Jira, freeze SLA with policy code SLA-FRZ-03 | sla-watch@dock | Marcus Webb | 2026-05-30 09:41 |
| FD-48319 (Tier-3, 11h left) | Three reopens, sentiment dropping | Route to retention specialist, attach account history | sentiment-scan@dock | Priya Anand | 2026-05-30 10:05 |
The ticket ID is a live pointer. Open the row and Dock pulls the current Freshdesk state through a fresh API read, so an approver never acts on stale status.
One workflow, end to end
Freshdesk fires a webhook the moment a Tier-1 ticket crosses 75% of its SLA window. The agent (sla-watch@dock, its own identity, not a borrowed seat) reads the ticket, the conversation, and the account tier through the Freshdesk API. It drafts a one-paragraph risk read and a recommended path. That draft lands as a Dock row routed to whichever support-ops lead is on shift. The lead approves, edits, or rejects in Dock. On approval, Dock writes the action back to Freshdesk (reassign, holding reply, SLA freeze) under the approver's name plus the agent's identity. The row stays. Six weeks later the QA team filters by approver, by policy code, by drafting agent.
Why this matters
Forrester found US CX quality fell to an all-time low across 2024, the third consecutive year of decline, with chatbot-led digital experience cited as a contributor (Forrester, 2024 US CX Index). Bolting an AI agent onto a ticket queue without an attributed approval layer is how that number gets worse. Freshdesk's own pitch for Freddy AI Copilot is agent productivity through summaries and reply suggestions (Freshworks, Freshdesk), which is the right capability and the wrong endpoint if the draft never meets a named reviewer. Dock makes the reviewer mandatory and the trail searchable. The same pattern that governs the broader support function governs SLA work specifically, and it matches how to run customer support with AI without losing the human signoff. The audit trail is the same one compliance teams ask for, and it depends on the agent having its own identity rather than logging in as a person.
This is the same shape used by Dock for IT operations for incident routing. Same architecture. Different platform. The audit posture holds across both.
Try it
Point sla-watch@dock at one Freshdesk view this week. Read the rows on Friday.
FAQ
Does Dock replace Freshdesk? No. Freshdesk holds tickets, conversations, and the SLA clock. Dock holds the agent's interpretation and the human approval. The ticket ID in each Dock row is a live pointer; current state comes from a fresh API read.
What if Freddy AI already drafts the reply? Freddy's draft is an input the agent reads. Dock records which draft was used, which support-ops lead approved it, and what edits were made before it reached the customer.
Who can approve an SLA freeze or reassignment? Whoever the support-ops lead designates per shift. Dock enforces the routing rule and records the named approver. Bulk approval is not a path the agent can take.
How long is the trail retained? Each row is permanent in the Dock workspace. Approver, agent identity, policy code, and timestamp persist for QA review and external audit without separate exports.
