Pando

Three routes are documented: pre-built connectors, named systems and authenticated action. The Technology Platform page states the platform integrates enterprise tools, carriers and third-party apps through pre-built connectors and APIs, over a proprietary middleware with an ETL pipeline giving continuous two-way data flow. The privacy policy's integrations disclosure names the third-party systems the platform connects with: ERP systems SAP and Oracle, logistics platforms DHL and FedEx, and compliance tools SAP GTS and Amber Road.

For action, the Pi release documents agents auditing and paying freight invoices across modes, services and currencies, executing claims and resolving disputes directly with carriers, and the audit specialist collaborates with carriers through the vendor's freight payment portal. No connector catalog page is published, the systems are named in the privacy policy rather than on a product page, and the number of connectors is not stated.

SourcePando, pando.ai/product/technology-platform, pando.ai/privacy-policy section 5, Pi release 2025-02-10read 2026-09-08

The Pi freight audit specialist runs an end-to-end multi-step process without a person moving it along: capturing invoice PDFs on arrival, reconciling charges against shipment data and live rates, flagging overcharges and duplicates, collaborating with carriers on disputes and routing cleared invoices for payment.

The Pi release documents procurement agents determining modal mix, optimizing carrier allocations, bundling lanes, negotiating rates and finalizing contracts. The Technology Platform page adds the customer-facing side: a no-code and low-code builder where workflows are configured visually with drag-and-drop across apps, with pre-built templates and rules and real-time alerts. That is vendor-run multi-step execution plus a customer workflow builder.

SourcePando, pando.ai/product/technology-platform, freight audit specialist page 2026-03-24 and Pi release 2025-02-10read 2026-09-08

The Technology Platform page documents a data platform that builds a knowledge graph across structured and unstructured data, normalizing master, transactional and third-party data while embedding workflows, constraints and network insights, with scalable storage and automatic anomaly correction.

The Pi release describes the same layer as an enterprise-specific supply chain knowledge graph, a real-time representation of the customer's logistics network that the agents reason over. That is a maintained, persistent, queryable structure over the customer's own operations.

No retrieval or index mechanism is described, and the enterprise-specific Logistics Language Models the release says are trained on it are the vendor's own trademark rather than a documented mechanism.

SourcePando, pando.ai/product/technology-platform and Pi release 2025-02-10read 2026-09-08

The freight audit page documents rate card updates landing behind an explicit one-click approve step before the rate master changes, and the benchmarking agent alerting the procurement team for review when rates exceed the expected threshold rather than acting alone.

That is the vendor's own review-and-approve surface, gating a write into the customer's rate master. The Technology Platform page adds access management and a single pane of glass across the procure-to-pay cycle. The approve step is documented for the rate card path, and no general autonomy configuration is described for the other agents.

SourcePando, freight audit page 2026-03-24 held and pando.ai/product/technology-platform read todayread 2026-09-08

The privacy policy's security section names role-based access and multi-factor authentication as implemented controls, alongside encryption in transit and at rest, continuous monitoring, penetration testing, vulnerability assessment, vendor security audits and an incident response procedure, with security logs retained for twelve months; the Technology Platform page adds access controls, threat detection and multi-AZ resilience. That is an access surface the buyer operates.

Compliance is hedged throughout and never reaches an attestation. The policy states compliance with ISO/IEC 27001 and ISO 28000 and describes ISO 27001- and ISO 28000-compliant controls, and the platform page says the platform adheres to ISO, SOC and CSA standards, with no certificate, auditor, report or date named anywhere and no trust portal on the site.

SourcePando, pando.ai/privacy-policy sections 1 and 8 and pando.ai/product/technology-platformread 2026-09-08

No agent run trace, decision log or audit record of what an agent did is documented. Three nearby features are not agent audit records. Security logs retained for twelve months in the privacy policy are infrastructure logging, not agent audit. Context captured as metadata for continuous improvement in the workflow builder is configuration data. Agents surfacing dollar impact on individual decisions is a result shown to the user rather than a retained record of the reasoning.

SourcePando, pando.ai/product/technology-platform and /privacy-policy section 7read 2026-09-08

The Technology Platform page uses the words "persistent memory" twice.

The integrations layer captures a persistent memory of all data exchanges for comprehensive analysis, and the networked collaboration layer keeps a persistent memory of carrier data and context to streamline communication with carriers, freight forwarders and customs agents across the procure-to-pay cycle; the no-code workflow layer captures context as metadata for continuous improvement.

That is state carried between interactions rather than assembled per run. No scope is stated (per user, per carrier relationship or per workflow), no lifetime or retention is given for the memory itself, and the phrasing sits close to a log of exchanges, which would be an application data store rather than agent memory.

SourcePando, pando.ai/product/technology-platformread 2026-09-08

Nothing is offered for a buyer with a residency requirement to select. The privacy policy discloses cloud hosting providers only as a category, names no hosting region and makes no residency commitment, and its international-transfer section relies on SCCs, adequacy decisions and binding corporate rules, which are legal cover rather than residency. The Technology Platform page's multi-AZ resilience is an availability property, not a region choice, and no self-hosted, private-cloud or customer-tenant option appears anywhere on the site.

SourcePando, pando.ai/privacy-policy sections 5 and 6 and /product/technology-platformread 2026-09-08

Pi has three named specialist teams, freight procurement, dispatch planning, and freight audit and payment, sold under a "Logistics Services as Software" model and stated to be deployed at large brands, and the Technology Platform page adds pre-built workflow templates and rules in the no-code builder.

Each team is a whole product a customer could adopt alone. No agent catalog or template library page exists on the current site, and the product navigation carries only the two TMS products and the Technology Platform, with no AI-agents page at all, so a buyer cannot browse and select among units.

SourcePando, Pi release 2025-02-10 and pando.ai/product/technology-platform; pando.ai/company/press-release/pando-launches-pi-ai-teams-for-logistics-enabling-autonomous-freight-procurement-planning-and-payments-for-global-brandsread 2026-09-08

Agents fire on event arrival rather than on request. The freight audit page documents agents capturing invoice PDFs as they arrive and auto-extracting every rate card that lands in the inbox, with a rate benchmarking agent running continuously against contract terms; the Technology Platform page adds real-time alerts and notifications configured in the workflow builder, and continuous data flow across the connected ecosystem through the middleware. No webhook or external channel surface is documented.

SourcePando, pando.ai/product/technology-platform read today, freight audit page 2026-03-24 and agentic walkthrough 2025-11-18 heldread 2026-09-08

No third-party provider, model selection surface, routing control or bring-your-own-key option is published. The Pi release states the knowledge graph continuously trains enterprise-specific Logistics Language Models, the vendor's own trademarked proprietary models. So the models are the vendor's own, and the customer has no choice among them.

SourcePando, Pi release 2025-02-10 and pando.ai/product/technology-platform; pando.ai/company/press-release/pando-launches-pi-ai-teams-for-logistics-enabling-autonomous-freight-procurement-planning-and-payments-for-global-brandsread 2026-09-08

API documentation lives at tms-docs.pando.ai, linked from the footer of every page under the label "APIs", and the Technology Platform page states the platform integrates enterprise tools, carriers and third-party apps via pre-built connectors and APIs over a proprietary middleware and ETL pipeline. The docs are public, with no login, under the title "Pando TMS API Docs". The reference body renders client side, so the endpoint list and the read and write scope are unconfirmed.

SourcePando, tms-docs.pando.ai probed and pando.ai/product/technology-platformread 2026-09-08

No evaluation of agent behavior is documented on the site, and no deployment-validation gate either: nothing describes a configuration or readiness check. Automatic anomaly correction on the data platform is data quality, and the headline savings, service-level and ROI figures are outcome marketing about the customer's freight spend rather than measurement of the agents.

SourcePando, pando.ai/product/technology-platform and Pi release 2025-02-10read 2026-09-08

Agents reach outside systems through connectors, APIs and the ETL middleware, and carrier follow-up runs through Pando's own freight payment portal, which is integration rather than computer use. No browser, desktop or remote control of an interface the vendor does not own is documented.

SourcePando, pando.ai/product/technology-platform and agentic walkthrough 2025-11-18read 2026-09-08